The Attenuation of Diabetic Nephropathy by Annexin A1 via Regulation of Lipid Metabolism Through the AMPK/PPARα/CPT1b Pathway

Diabetic nephropathy (DN) is one of the most serious complications of diabetes affecting many people worldwide. Despite optimal conventional therapy, a significant proportion of patients with DN still progress to end-stage renal disease. Thus, novel approaches to preventing or slowing the progression of DN are urgently needed.

Inflammation and abnormal metabolism play important roles in the pathogenesis of DN. Increased infiltration of proinflammatory macrophages and release of proinflammatory cytokines contribute to insulin resistance and generation of advanced glycation end products (1). Correspondingly, metabolic reprogramming exacerbates the inflammatory process via mitochondrial dysfunction (2). Intervention in this process may attenuate renal injuries in diabetes.

Annexin A1 (ANXA1), a 37-kDa protein, is a member of the annexin superfamily. It binds to cellular membranes in a calcium-dependent manner (3). ANXA1 participates in various cellular processes (e.g., cell proliferation, differentiation, and apoptosis) (4). To date, studies on its protective mechanisms have mainly focused on its proresolving effects, including effects on innate immunity, cytokine expression, and T-cell activation (5). Additionally, ANXA1 is thought to modulate metabolism (6). For instance, plasma level of ANXA1 correlated with fatty liver index as well as plasma cholesterol level in patients with type 2 diabetes, indicating a link between lipid handling and ANXA1 (7). ANXA1 replenishment may improve the function and survival of isolated islets in vitro and improve islet transplantation in a mouse model of diabetes (8,9). Furthermore, Anxa1 knockout (KO) mice were more susceptible to weight gain and diet-induced insulin resistance and had higher adipose tissue lipolysis (10). ANXA1 has been shown to protect kidneys from functional decline in type 1 and type 2 diabetes animal models (7,11). However, the protective mechanisms of ANXA1 in DN have not been well characterized.

In this study, we reported that genetic deletion of ANXA1 in DN mice aggravated renal injuries. Transcriptomics and lipidomics analyses revealed that ANXA1 deletion altered lipid metabolism profiles. Ac2-26, an ANXA1 mimetic peptide, had a therapeutic effect against lipid toxicity in diabetic mice. In vitro analyses showed that ANXA1 inhibition downregulated mitochondrial fatty acid oxidation (FAO) activity by suppressing peroxisome proliferator–activated receptor-α (PPARα) and the downstream carnitine palmitoyltransferase 1b (CPT1b), resulting in intracellular lipid accumulation and injuries in HK-2 cells. In addition, there was a significant association between ANXA1 and lipid accumulation in patients with DN. Taken together, this study provides the first evidence that the ANXA1-FPR2/ALX-AMPK-PPARα-CPT1b axis influences mitochondrial lipid metabolism. It may offer a novel viewpoint that ANXA1, as a well-recognized proresolving protein, could regulate cellular lipid metabolism to ameliorate disease progression in DN.

Male C57BL/6 Anxa1 KO (Anxa1^−/−) mice were purchased from Cyagen Biosciences Inc. (Jiangsu, China) (12). The founders were genotyped by PCR, and mutation was confirmed by DNA sequencing. The presence of targeted genes was determined by PCR of tail DNA. Ethical approval (license number LA2011-046) for animal studies was granted by the Ethics Committee of Peking University Health Science Center. All procedures adhered to institutional guidelines on animal experimentation.

HK-2, human immortalized proximal tubule epithelial cells (PTECs), were cultured in DMEM/nutrient mixture F-12 and enriched with 1% penicillin-streptomycin and 10% FBS at 37°C with 5% CO₂. The conditionally immortalized human podocyte cell line was provided by Prof. Jochen Reiser (Rush University, Chicago, IL) and cultured as previously described (13).

Thirty patients with biopsy-proven DN diagnosed at Peking University First Hospital were recruited into this study, as described previously (14). Control kidney samples (n = 7) were obtained from healthy kidney poles of individuals without diabetes or kidney disease receiving tumor nephrectomies. The study complied with the Declaration of Helsinki and was approved by the Ethics Committee of Peking University First Hospital (No. 2017[1280]).

Diabetic mice were induced by feeding high-fat diet (HFD) (HfkBio; 60% fat) for 1 month followed by an intraperitoneal injection of 50 mg/kg streptozotocin (STZ) (Sigma-Aldrich). Diabetic mice were then maintained on HFD for 20 weeks. Ten days after STZ injection, random blood glucose was measured, and mice with blood glucose >16.7 mmol/L were induced successfully. As the control, nondiabetic mice were fed with standard-fat diet (10% fat) and treated with 0.1 mol/L sodium citrate. Blood glucose levels were monitored weekly throughout the study duration.

The protocols of Ac2-26 treatment were described previously (14). Briefly, for the treatment study in diabetic Anxa1 KO mice, all four groups of mice were induced DN with HFD/STZ. One group of diabetic Anxa1 KO mice and diabetic wild-type (WT) mice were maintained on Ac2-26 treatment (2 mg/kg, in PBS), while the other two groups of mice were treated with the vehicle (PBS) for 16 weeks. For the treatment study in db/db mice, 10-week-old male db/db mice were maintained on Ac2-26 (2 mg/kg, in PBS) or the vehicle (PBS) treatment for 10 weeks. Age- and sex-matched db/m mice treated with the vehicle served as a nondiabetic control group.

Blood glucose, triglyceride, and cholesterol concentrations were analyzed with commercial kits (Biosino Bio-technology & Science, Inc., Beijing, China). Urine albumin concentrations were measured with a kit from Bethyl Laboratories. Urine creatinine levels were measured with the Creatinine Companion kit (Exocell). Urine albumin levels were standardized to urine creatinine excretion and documented as urinary albumin-to-creatinine ratio (uACR).

Staining of sections of the kidney (4 µm) was done using periodic acid Schiff (PAS). Glomerular mesangial matrix expansion, sclerosis, and tubulointerstitial injury were assessed as described previously (15–17).

Glomerular area was measured by tracing around the perimeter of the glomerular tuft. The mesangial matrix expansion area was assessed from the digital photographs of glomeruli and presented as a proportion of PAS-stained per glomerular cross-sectional area. The tubulointerstitial injury index was determined by assessing the extent and severity of tubular dilation, atrophy, and loss of tubular cells. We imaged 20 views of a kidney section (magnification ×200) and scored as follows: 0 for no injury, 1 for <25%, 2 for 25–50%, 3 for 50–75%, and 4 for >75% tubulointerstitial injury (18). Quantitation analyses were done on Image-Pro Plus.

Paraffin sections (4 µm) were immunostained with antibodies against adipose differentiation-related protein (ADRP) (Abcam), F4/80 (Serotec), α-smooth muscle actin (Abcam), PPARα (Abcam), CPT1b (Abcam), or phosphorylated (phospho-) Thr¹⁷² AMPK (Cell Signaling Technology). After incubating with primary antibody overnight at 4°C, the membranes were incubated with peroxidase-conjugated secondary antibodies (ZSGB-BIO, Beijing, China), and the signal was developed using 3,3'-diaminobenzidine tetrahydrochloride (ZSGB-BIO). Sections were then examined under a light microscope and digitized with a high-resolution camera.

Electron microscopic sample handling and detection were performed by the Electron Microscopic Central Laboratory of Peking University First Hospital, as described previously (19). Image-Pro Plus was used to measure glomerular basement membrane (GBM) thickness, as well as foot process width (20). All glomeruli were selected, and 15 electron micrographs were taken in each glomerulus.

To analyze tubular mitochondrial morphology, the tubulointerstitium was photographed at ×7,000 magnification. Based on published methods (21), aspect ratio was defined as the ratio of mitochondrial major and minor axes. The degree of form factor was defined as perimeter² / (4π × surface area). The morphology of at least 120 mitochondria was determined for each condition.

RNA sequencing (RNA-seq) analysis was outsourced to OE Biotech (Shanghai, China). Briefly, total RNA was isolated, and the integrity of the RNA was determined on an Agilent 2100 bioanalyzer. Using the TruSeq Stranded mRNA LTSample Prep Kit (Illumina), we processed libraries, which were sequenced on the Illumina HiSeq X Ten. Differentially expressed genes (DEGs; false discovery rate–adjusted P < 0.05 and fold change >2 or fold change <0.5) were determined and then analyzed for Kyoto Encyclopedia of Genes and Genomes pathway. RNA-seq data are deposited in the National Center for Biotechnology Information Sequence Read Archive under accession number PRJNA688802.

Lipids were isolated from tissues according to the modified Bligh and Dyer’s isolation approach and dried using SpeedVac under OH mode (22). We resuspended the lipid extracts in chloroform/methanol 1:1 (v/v) spiked with indicated internal standards. We conducted all lipidomic assessments on an Exion UPLC system coupled with a QTRAP 6500 PLUS system (Sciex) as previously documented (22,23). All quantification experiments used internal standard calibration (24,25).

RNA was isolated from kidney cortex or cultured cells using TRIzol reagent and reversed into cDNA with High-Capacity cDNA Reverse Transcription Kits (Applied Biosystems) following the manufacturer’s instruction. Gene expression was standardized relative to 18S ribosomal RNA or β-actin. Primers are indicated in Supplementary Table 1.

At 80% confluence, the cells were stimulated for 24 h with high glucose and palmitic acid (HGPA) at a final concentration of 30 mmol/L glucose and 200 μmol/L palmitate (26,27). The cells were simultaneously exposed to 100 μmol/L A769662, an AMPK agonist (Abcam). Cells were pretreated with 10 μmol/L WRW4 (Meilun Bio), an ALX/FPR2 antagonist, before being treated with 10 nmol/L human recombinant ANXA1 (hrANXA1) (R&D Systems) (28). Lipids were stained with an Oil Red O stain kit (Solarbio Life Sciences).

Anti-human ANXA1 shRNA (5'-CCAGCGCAATTTGATGCTGAT-3') as well as the negative control shRNA (5'-TTCTCCGAACGTGTCACGT-3') were developed, synthesized, and then cloned into the pFU-GW-016 vector by GeneChem Co. (Shanghai, China). Stably ANXA1-silenced HK-2 cells were established using lentiviral constructs expressing shRNA against ANXA1 (shANXA1), and the negative controls (shCtrl) were applied to parallel cultures. Cells were kept in the media containing 2 μg/mL puromycin (Sigma-Aldrich).

Podocyte cells were transfected with siRNAs targeting human ANXA1 (HSS100503; Invitrogen) using X-tremeGENE siRNA Transfection Reagent (Roche Applied Science) according to the manufacturer’s instructions. For human FPR2/ALX knockdown in HK-2 cells, FPR2/ALX siRNA (50 nmol/L) (RiboBio Co., Ltd.) was transfected as the experimental group, and the corresponding negative control siRNA was transfected as the control group.

Western blot analysis was performed as previously described (29). The membranes were incubated with antibodies against ANXA1 (Abcam), phospho-Thr¹⁷² AMPK (Cell Signaling Technology), total AMPK (Cell Signaling Technology), PPARα (Abcam), CPT1b (Proteintech), GAPDH (Abcam), or β-actin (Abcam).

Mitochondria were visualized using MitoTracker (Beyotime) according to the manufacturer’s instructions. Mitochondrial morphology was examined in 15 random fields/sample. The percentage of cells with an altered mitochondrial pattern, in terms of fragmentation and perinuclear redistribution, on total cells per field was semiquantitatively analyzed (30). Analysis of mitochondrial network morphology was done using the Mitochondrial Network Analysis toolset on Fiji (ImageJ) and standardized to the number of Hoechst-positive cells (31). Two-dimensional microscopy images were minimally processed using the “unsharp mask” to enhance the sharpness and then binarized (namely, skeletonizing followed by measurement of individuals, networks, mitochondrial footprint, and mean branches per network).

This process was performed using TUNEL Apoptosis Assay Kit (Beyotime) according to the manufacturer’s instructions. Ten random fields of cells were counted to determine the percentage of cells undergoing apoptosis under high-power magnification.

Mitochondrial FAO and function were explored by a Seahorse flux analyzer (Agilent Technologies) according to the manufacturer’s instructions. Briefly, 2 × 10⁵ cells/well were seeded into XF24 cell culture microplates and cultured in growth medium (Agilent Technologies). Before the assay, the cells were rinsed with assay medium. For FAO assay, palmitate/BSA or BSA was introduced into appropriate wells. Oligomycin (2 mmol/L), carbonyl cyanide p-trifluoromethoxyphenylhydrazone (1 mmol/L), and rotenone/antimycin A (0.5 mmol/L) were sequentially added into the culture, and oxygen consumption rate (OCR) was assessed.

Statistical analysis was done by SPSS version 22.0 (IBM). Results were presented as mean ± SD. Student t test or one-way ANOVA with Bonferroni correction was used as appropriate. Correlation was examined using the Pearson test for bivariate normal distribution; otherwise, Spearman test was used as appropriate. P < 0.05 indicated statistical significance.

The data sets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.

To assess its function in DN, ANXA1 was deleted in the C57BL/6 background to generate Anxa1^−/− mice. Anxa1^−/− mice, as well as their WT littermates, were treated with HFD plus STZ to induce DN. Their metabolic characteristics were recorded (Supplementary Table 2). There was an increase in kidney weight and kidney/body weight in diabetic Anxa1^−/− mice compared with diabetic WT mice, although the difference did not reach the statistical level. In addition, diabetic Anxa1^−/− mice exhibited significantly higher levels of uACR (Fig. 1A). PAS analysis showed that diabetic Anxa1^−/− mice also manifested more severe histological changes in kidneys, including glomerular size, mesangial matrix expansion, and tubulointerstitial lesions (Fig. 1B–E). To further evaluate the tubulointerstitial injuries, tubulointerstitial F4/80-positive cells, tubulointerstitial fibrosis, and the expression of the tubular injury markers, including kidney injury molecule-1 (KIM-1) and neutrophil gelatinase-associated lipocalin (NGAL), were measured in kidneys. The results showed significantly higher levels of macrophage infiltration and α-smooth muscle actin expression in tubulointerstitium and significantly higher mRNA levels of the tubular injury markers Kim-1 and Ngal in diabetic Anxa1^−/− mice than in diabetic WT mice (Supplementary Fig. 1). Transmission electron microscopy (TEM) analysis showed that GBM thickness and foot process width were significantly greater in diabetic Anxa1^−/− mice than in diabetic WT mice (Fig. 1F–H). Overall, ANXA1 deficiency exacerbated kidney injuries in diabetes.

To investigate how Anxa1 deficiency exacerbated kidney injuries in diabetes, we performed RNA-seq analysis of the renal cortex. Volcano plot demonstrated that most DEGs were downregulated in kidneys of diabetic Anxa1^−/− mice as compared with diabetic WT mice (Supplementary Fig. 2A). Specifically, 239 DEGs were identified, with 200 downregulated and 39 upregulated. Kyoto Encyclopedia of Genes and Genomes enrichment analysis showed that, compared with diabetic WT mice, diabetic Anxa1^−/− mice were enriched for pathways associated with cellular metabolism, including regulation of lipolysis in adipocytes, glycolysis/gluconeogenesis, and tyrosine metabolism (Supplementary Fig. 2B). Moreover, gene set enrichment analysis showed that downregulated genes in diabetic Anxa1^−/− mice were significantly enriched in AMPK signaling, PPAR signaling, and fatty acid degradation (Fig. 2A–C). Interestingly, the enriched pathways are associated with diabetes and related metabolic disorders and are essential for regulating energy metabolism (32,33). Heat map analysis found that the contents of genes linked to fatty acid metabolism were mostly downregulated in diabetic Anxa1^−/− mice (Fig. 2D and Supplementary Table 3). Transcriptomic data were confirmed using real-time quantitative PCR analysis of select genes, including Cpt1b, Pparα, Fabp4, and Hsl (Fig. 2E). Moreover, immunohistochemistry confirmed that phospho-AMPK, PPARα, and CPT1b expression was markedly lower in the diabetic Anxa1^−/− mice compared with diabetic WT mice (Supplementary Fig. 2C).

Analysis of kidney lipid profiles by lipidomics revealed that Anxa1 deficiency significantly increased the levels of various triacylglycerol (TAG) species in STZ/HFD-induced DN (Fig. 3A and Supplementary Table 4). Consistently, TEM showed that diabetic Anxa1^−/− mice had larger lipid droplets in PTECs compared with diabetic WT mice (Fig. 3B and C). Furthermore, mitochondrial structure was severely damaged in diabetic Anxa1^−/− mice, in which proximal tubular mitochondria were disorganized and swollen (Fig. 3B and D). Taken together, these in vivo results suggest that ANXA1 depletion promotes intrarenal lipid accumulation and drives mitochondrial impairment in DN mice model.

In our recent study, we found that Ac2-26, an ANXA1 mimetic peptide, had the therapeutic potential for alleviating kidney injuries in diabetic mice (14). We further investigate whether Ac2-26 could attenuate lipid toxicity in diabetic Anxa1^−/− mice and db/db mice. After Ac2-26 treatment, the deposition of lipid droplets was significantly decreased, and the mitochondrial morphology was improved, compared with vehicle-treated diabetic mice (Supplementary Fig. 3A and B). In addition, Ac2-26 treatment mice showed decreased levels of profibrotic factors and the tubular injury marker as compared with vehicle-treated diabetic mice (Supplementary Fig. 3C and D).

To elucidate the mechanism of ANXA1 in lipid metabolism of the kidney, especially in PTECs, which mainly use fatty acids as the energy source, we silenced ANXA1 in HK-2 cells using a lentiviral system (Fig. 4A). Oil Red O staining revealed that shANXA1 group treated with HGPA had significantly increased lipid accumulation (Fig. 4B). Mitochondrial morphology analysis showed that the shCtrl group without HGPA treatment had the filamentous network of elongated mitochondria. Massive mitochondrial fragmentation as well as redistribution of these small and round organelles in the perinuclear site of cells occurred upon ANXA1 knockdown (Fig. 4C). In addition, mitochondrial individuals, networks, footprints, and mean branches were significantly decreased in the shANXA1 group treated with HGPA compared with the shCtrl group treated with HGPA (Fig. 4D).

Metabolic derangement and ultrastructural mitochondrial alterations were observed in the shANXA1 group treated with HGPA. Hence, to evaluate the metabolic pattern of PTECs, we assessed OCR of HK-2 cells with or without HGPA treatment. We found that HGPA treatment suppressed the spare respiratory capacity and ATP production (Supplementary Fig. 4A). Additionally, ANXA1 knockdown significantly reduced basal respiration, ATP production, and spare respiratory capacity when cells were treated with HGPA (Fig. 4E), indicating that mitochondrial respiration is severely impaired. Consistently, exogenous supplementation of hrANXA1 improved the mitochondrial respiratory function of HGPA-treated HK-2 cells (Supplementary Fig. 4B). Evaluation of fatty acid utilization using the Seahorse bioanalyzer revealed that FAO activity was markedly reduced in the shANXA1 group treated with HGPA, as compared with the shCtrl group treated with HGPA (Fig. 4F). Moreover, FAO activity was improved by hrANXA1 supplementation in HGPA-treated HK-2 cells (Supplementary Fig. 4C). However, ANXA1 knockdown did not obviously affect the glycolysis in HK-2 cells (Supplementary Fig. 4D).

Given that lipotoxicity causes kidney cell death, we assessed apoptosis in HK-2 with or without ANXA1 knockdown. TUNEL analysis revealed more apoptotic cells in the shANXA1 group treated with HGPA compared with the shCtrl group treated with HGPA (Fig. 4G). Moreover, HGPA-triggered expression of proinflammatory and profibrotic factors, including collagen type IV α 1 (COL4A1), monocyte chemotactic protein 1 (MCP-1), and tumor necrosis factor (TNF), was enhanced by ANXA1 silencing (Fig. 4H).

In addition, since lipid dysmetabolism in podocytes also plays a pathogenic role in DN, to determine the role of ANXA1 in HGPA-induced podocyte injury, we knocked down ANXA1 by siRNA in podocytes prior to HGPA treatment. We found that siANXA1 in podocytes treated with HGPA could significantly increase lipid accumulation, apoptosis, and the expression of proinflammatory and profibrotic factors as compared with the siCtrl group treated with HGPA (Supplementary Fig. 5).

Both the mRNA and protein levels of PPARα and CPT1b were significantly decreased in vitro in the shANXA1 group treated with HGPA as compared with the shCtrl group treated with HGPA (Fig. 5A and B). These results were consistent with the above-mentioned in vivo findings. Consistently, supplementation of hrANXA1 restored the expression of PPARα and CPT1b in the shANXA1 group treated with HGPA (Fig. 5C). However, the regulatory effects of hrANXA1 were attenuated in the presence of WRW4 (ALX/FPR2 antagonist) as well as upon FPR2 knockdown, indicating ANXA1 stimulates PPARα/CPT1b signaling via FPR2/ALX (Fig. 5C and Supplementary Fig. 6). The activity of AMPK, the main energy-sensing enzyme, was suppressed in kidneys of subjects with diabetes (34,35). Western blot analysis revealed significantly reduced phospho-AMPK level in the shANXA1 group treated with HGPA compared with controls, which was consistent with the abovementioned immunohistochemical data from diabetic Anxa1^−/− mice. To determine whether ANXA1 influences the AMPK-mediated signaling pathway, we treated shANXA1-transfected HK-2 cells with A769662, an AMPK agonist. It was found that A769662 restored the protein expression of PPARα and CPT1b (Fig. 5D). The increased lipid accumulation observed in the shANXA1 group treated with HGPA was significantly ameliorated by A769662. Furthermore, the shANXA1 group treated with HGPA led to a marked increase in apoptosis, whereas A769662 partially prevented this increase (Fig. 5E and F). In addition, expression of COL4A1, MCP-1, and TNF were significantly increased in the shANXA1 group treated with HGPA, and A769662 was able to partially ameliorate the increase (Fig. 5G).

Taken together, these data demonstrated that ANXA1 knockdown suppressed the AMPK-PPARα-CPT1b signaling pathway via FPR2/ALX and thereby inhibited mitochondrial FAO activity and exacerbated lipotoxicity-induced inflammation and fibrosis in PTECs. Defective FAO and the subsequent inflammation, fibrosis, and apoptosis contribute to the progression of DN. ANXA1-mediated improvement of FAO, inflammation, fibrosis, and apoptosis in PTECs may partly halt this vicious cycle (Fig. 5H).

We have recently reported increased intrarenal expression of ANXA1 (or ANXA1 mRNA) in patients with DN compared with healthy control subjects. There was a significant negative correlation between the level of ANXA1 (or ANXA1 mRNA) in kidneys and renal function. ANXA1 was mainly expressed in tubular epithelial cells and podocytes in kidneys of patients with DN (14). In this study, we further found that renal biopsies of patients with DN exhibited more intense ADRP staining than healthy control subjects, and ADRP significantly correlated with estimated glomerular filtration rate (r = −0.651; P < 0.001). Moreover, tubulointerstitial ANXA1 expression significantly correlated with ADRP (r = 0.399; P = 0.029) (Supplementary Fig. 7). Taken together, these data from human renal biopsy were consistent with the abovementioned in vitro and in vivo data, indicating that ANXA1 was associated with lipid abnormalities in patients with DN.

Recent evidence has shown the important role of mitochondrial dysfunction, especially impaired mitochondrial FAO, in the development and progression of DN (35). In this study, we demonstrated a protective role for ANXA1 in HFD/STZ-induced DN mice. More importantly, we uncovered a novel mechanism that ANXA1 may improve mitochondrial FAO via AMPK/PPARα/CPT1b signaling pathway in PTECs, by which it may reduce intracellular lipid accumulation and ameliorate lipotoxicity-mediated tubular damage in DN.

We constructed Anxa1 KO mice and developed a DN model using HFD and STZ and then assessed the role of ANXA1 in DN. Consistent with previous studies using diabetic mice models (7,11), we found that Anxa1 KO worsened diabetes-induced kidney injuries, confirming that ANXA1 could mitigate established DN. In particular, Anxa1 KO aggravated tubular cell injuries, including tubular lipid accumulation, inflammation, and fibrosis, which may trigger glomerular damage via the cross talk between tubular cells and glomerular cells in DN mice (36,37). Transcriptomic analysis showed that genes associated with fatty acid degradation and PPAR signaling were suppressed in diabetic Anxa1^−/− mice compared with diabetic WT mice. PPARα, abundantly expressed in proximal tubular cells in kidneys, is an important regulator of fatty acid metabolism. PPARα activation could upregulate gene expression involved in fatty acid transport, binding, and activation (38–41). The rate-limiting enzyme in fatty acid β-oxidation, CPT1, is also induced by PPARα (42). Intriguingly, DEG analysis showed that several lipid oxidation– and lipolysis-related genes, including PPARα and CPT1b, were downregulated in diabetic Anxa1^−/− mice compared with diabetic WT mice. In addition, lipidomic analysis showed that diabetic Anxa1^−/− mice had a significantly higher abundance of TAG. Thus, after integration of transcriptomics and lipidomics as well as renal pathological analysis, we proposed that ANXA1 deletion altered fatty acid metabolism, contributing to intrarenal lipid accumulation and the corresponding lipotoxicity in DN.

PTECs have a high level of baseline energy consumption and rely mainly on FAO as their energy source, making PTECs vulnerable to an altered lipid environment. In addition, we found that in patients with DN, ANXA1 expression was markedly increased in the tubulointerstitium, and the tubulointerstitial ANXA1 expression exhibited better correlations with renal function and lipid deposition in the kidney than glomerular ANXA1 expression. Thus, we considered that ANXA1 may play a greater functional role in renal tubular cells than in glomerular cells. In this study, we found that mitochondrial function and FAO activity were impaired when ANXA1 gene was silenced in PTECs. However, Seahorse analysis showed that ANXA1 knockdown did not obviously affect glycolysis, highlighting its importance in FAO regulation. The pathogenic role of lipotoxicity has been generally proposed to be an overload in intracellular free fatty acid, leading to an accumulated triglyceride pool (43). We considered that this significant alteration of fatty acid metabolism in tubular cells causes lipid accumulation and lipotoxicity upon ANXA1 knockdown. This lipotoxicity, in turn, enhances a vicious cycle, leading to the exacerbation of abnormal metabolism. In patients with diabetes and mice models, activation of AMPK was reduced in kidneys (44). Adiponectin and leptin are reported to induce PPARα gene expression and improve the subsequent FAO via AMPK activation (45,46). Interestingly, ANXA1 could promote anti-inflammatory macrophage phenotype through the FPR2/CaMK/AMPK pathway in skeletal muscle injury regeneration (47). We further extended the results by demonstrating that ANXA1 regulates FPR2/PPARα/CPT1b and the downstream lipotoxicity by stimulating AMPK activation in proximal tubule epithelial cells in DN. Collectively, our in vitro data suggest that ANXA1 depletion decreased FAO, probably via FPR2/ALX-AMPK-PPARα-CPT1b signaling pathway.

In addition to tubular cells, ANXA1 was also expressed in podocytes (14). We observed that knockdown of ANXA1 in podocytes treated with HGPA could aggravate lipid deposition and apoptosis, suggesting ANXA1 may also play a protective role in lipid metabolism of podocytes. However, the potential mechanism of ANXA1 in podocytes needs further investigation. In addition, considering the limitation of global Anxa1 KO and bulk tissue RNA-seq, further studies are needed to determine cell-type responses to ANXA1 depletion in DN.

Last but not least, we found that ANXA1 silencing enhanced proinflammatory and profibrotic cytokine expression. Increasing evidence indicates that changes of metabolic status in immune cells may affect inflammation (48,49). Consistently, metabolic reprogramming of tubular epithelial cells has been suggested to modulate profibrotic phenotypes (50). Thus, we considered that regulation of FAO activity might be one of the anti-inflammatory mechanisms of ANXA1.

In conclusion, our data provide evidence that ANXA1 mitigates kidney injuries in DN via regulating CPT1b and improving mitochondrial FAO, offering a potential therapeutic opportunity. Further studies are required to better understand the molecular pathways regulated by ANXA1 in DN.

Funding. This study was funded by grants from the National Key Research and Development Program (2016YFC1305405), grants from the National Natural Science Fund (82090020, 82090021, 82070748, 91639108, 81770272 and 81970425), a grant from the CAMS Innovation Fund for Medical Sciences (2019-I2M-5-046), and a grant from the University of Michigan Health System and the Peking University Health Sciences Center Joint Institute for Translational and Clinical Research (BMU2017JI001).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. L.W. performed the majority of experiments and drafted and revised the manuscript. C.L. produced Anxa1^−/− mice. D.-Y.C., R.Z., and M.Z. designed the experiments and revised the manuscript. S.M.L. and G.S. performed lipidomic analysis. L.Z. and M.C. designed the research, revised the manuscript, and supervised the study. L.Z. and M.C. are the guarantors of this work and, as such, had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.